Anthropic's Claude Code hits 125K stars: why developers are skipping the IDE

I've been watching how developers actually integrate AI into their day, and a pattern keeps showing up: the IDE is no longer the center of gravity. People are running agents in the terminal, tagging them on PRs, and skipping the chat window.

Claude Code is the clearest expression of that. The GitHub metrics make it hard to argue it's a niche workflow anymore.

What Happened

anthropics/claude-code hit 125K stars and 20.5K forks with 53 contributors. The most recent CHANGELOG and feed.xml updates landed within the hour as I'm writing this. The repo went GA with v1.0.0 in May 2025, so most of that traction came in roughly twelve months.

What I'd flag from the repo state:

• The codebase is 47.1% Shell, 29.2% Python, 17.7% TypeScript. That distribution tells you Anthropic built this as a cross-platform CLI, not a Node app dressed up as one.
• 5K+ open issues and 544 open PRs. That's the kind of backlog you get when a user base scales faster than the maintainer team can triage.
• Commits co-authored by @claude regularly appear in the contributor list. Anthropic is dogfooding the tool on its own codebase, which is the most credible signal you can ship.
• The npm install path is now deprecated. Recommended installs go through curl | bash, irm | iex, Homebrew, or WinGet.

Key Features

• Terminal-native agent: runs in your shell, reads the codebase, and responds to natural language for edits, explanations, and git operations. The bet is that most developers want AI assistance in the terminal, not the IDE.
• Plugin system: the plugins/ directory ships with extensible commands and agents, including a code-review plugin that posts inline comments on PRs. The plugin model turns Claude Code from a CLI into a platform.
• IDE and GitHub integration: works in the terminal, the editor, or directly on GitHub by tagging @claude. Most teams underestimate the GitHub integration, since it pulls AI review into where collaboration already happens.
• DevContainer support: ships with a .devcontainer config that pre-loads the VS Code extension and configures firewall rules for marketplace access. Useful for teams standardizing remote dev environments.
• Native installers across platforms: curl and PowerShell scripts, Homebrew, WinGet. The move away from npm signals who Anthropic thinks the audience is.
• Documented privacy controls: limited retention for sensitive data, restricted session access, and a no-training-on-feedback policy.

Why It matters

A few things I'm seeing more broadly that line up with this:

• The IDE is losing ground as the default surface: Cursor and Copilot are still huge, but the developers I talk to who are most deeply involved in AI workflows are spending more time in the terminal. Claude Code's traction validates that.
• GitHub is becoming an agent runtime: Tagging @claude on an issue or PR turns the platform itself into the interface. That's a meaningful shift away from "open the AI app, paste the code, copy the response."
• The Node deprecation matters: When a project moves away from npm toward native installers, its audience isn't just JavaScript developers anymore. Claude Code wants to be your CLI tool, not your dependency.

Collapsing "ask the AI" and "apply the change" into a single step is where the productivity gain actually lies. Everything else is interface polish.

Example Use Case

A backend engineer is adding a new REST endpoint to a Django project: a list view, a detail view, serializers, URL routing and tests. The boilerplate is the slow part.

They run claude in the project root and type: "Add a REST endpoint for user notifications with list and detail views, serializers, and unit tests."

Claude Code reads the existing project structure, picks up the Django REST framework patterns already in use, generates the view, serializer, URL configuration, and test files, and then stages the changes in Git. The engineer reviews the diff, tweaks a queryset filter and commits.

This is the workflow I'd demo to anyone still copy-pasting between a chat window and their editor. The latency reduction alone is the sell.

Competitive Context

A few things stand out when you put Claude Code next to the obvious alternatives:

• Scope is narrower than Cursor or Copilot: The README is explicit about it being a terminal-based agent for coding workflows. That's a feature, not a limitation. Cursor wants to own the IDE; Claude Code wants to own the shell.
• 20.5K forks is the number to watch: Stars are easy. Forks mean people are actually building on it, customizing the plugin system, or adapting commands for their team. That's the difference between a popular repo and a foundational one.
• The plugin architecture creates a moat Copilot doesn't have: Once developers invest in writing custom commands and agents per repo, they lock in.

The competitive question isn't whether Claude Code is better than Cursor. It's whether developers want AI in the IDE or AI in the terminal. The fork count and contributor activity suggest a real audience is voting for the terminal.

My Take

The thing I keep coming back to: 125K stars is a great vanity metric, but 5K+ open issues and 544 PRs is the more interesting number. That's a tool people are pushing hard, hitting limits on, and trying to extend.

The npm deprecation is the sign that Anthropic is playing a longer game. They're optimizing to be the default CLI for AI-assisted coding across every stack, not specifically for the JavaScript developer audience.

What I'm watching next: whether the plugin ecosystem develops outside Anthropic. If teams start shipping their own plugins for internal tooling and workflows, Claude Code becomes a platform. If it stays official-plugins-only, it will remain a great CLI. Those are very different futures, and the next six months should tell us which one is happening.